He Banned AI From His Writing Group. His Own Cover Got Called AI Slop.

“AI slop” has stopped meaning anything. It’s what people say now when they don’t like a picture.

That isn’t my line. It came from the admin of a big fiction writing group on Facebook, a man who is vehemently against AI in his own work, and he learned it the hard way. Strangers looked at his book cover, decided a machine made it and said so in public. They were wrong. A human being painted it by hand, years before any of the image generators existed.

Then, in the same post where he told that story, he announced he’d be banning AI from the group, with moderators who would spot it by eye. Does anybody else see the problem?

Can a Facebook group owner ban AI content from the group?

Of course he can. It’s his group.

He started it eleven years ago, and he gets to decide what goes on the walls. He put it well himself: the moderators aren’t going out hunting people who use AI, they just won’t let it in the door. Fair enough. Plenty of groups ban fan fiction, poetry on the wrong day, self-promotion and links. A ban on AI covers and AI prose is one more house rule, and anybody who doesn’t like it can find another house.

The new rule gives one warning, and the second offense is a ban. It also covers comments written with AI. And he added something I agree with completely: members who call other people’s work “AI slop” when it isn’t, or who throw insults around in the comments, will be dealt with too.

I think the ban itself is silly. I use AI every day, I make my covers with it, and the people screaming about it have only gotten louder since I started saying so. But it’s his call. A writing group isn’t a public utility, and nobody’s forced to stay in it.

I don’t join writing groups these days. They were hostile places long before AI showed up, and the people against AI are only the latest bunch to be furious in them. Would I want to post my work in a room where the moderators guess? No. Would I tell the owner he can’t run his room that way? Also no.

The cover that was painted eight years too early

The admin wrote a time-travel novel in 2013 and published it in 2015. The cover was painted on canvas, in watercolor, in 2014, by a man he knew personally: his kids’ elementary school art teacher.

A few weeks ago he ran a Facebook ad for that book.

The comments filled up with people calling the cover AI slop.

Midjourney opened its public beta in July 2022. OpenAI dropped the waitlist for DALL-E that September. His cover was eight years old by then. Nobody making those comments checked a date, asked a question or looked at the brushwork. They looked, they guessed, and they announced the guess as a fact.

There wasn’t any other method.

His own conclusion was blunt. “AI slop” has become a generic insult for art somebody doesn’t like. He’s right, and it’s a rotten thing to do to any artist who did the work by hand.

How do Facebook group moderators decide a post used AI?

Mostly, they can’t. I’ve looked at plenty of covers, and I couldn’t tell you with certainty how half of them were made. Neither can you, and neither can a moderator.

Somebody in the thread asked exactly that: how will you determine it? Another member answered it straight. “By the vibes,” she said, and added that she hated it and wished the industry had a real test. One of the group’s own moderators said the admins have been critiquing writers for more than ten years and can trust their guts.

Trust their guts? A gut is exactly what called a 2014 watercolor AI slop. A guess with a moderator badge on it is still a guess.

The tools don’t settle it either. One member suggested running suspect images through SynthID, Google’s watermark checker. It finds a watermark when the tool that made the image put one there, and Google says its partners, including OpenAI, are adding it too. A clean result only means no watermark was found. It can’t prove a person painted anything. As I’ve written before, AI detection can’t be evidence. That goes for text detectors, and it goes double for eyeballs.

So the group now has two rules pointing in opposite directions. Members can’t accuse each other of AI without proof. Moderators can remove posts for AI with no proof at all. Who’s going to get hit when a moderator’s gut is wrong? The same kind of person who got hit on that ad: somebody who did the work by hand and can’t prove a negative.

Put yourself on the other end of that. You spent three weeks with a designer, you paid for the cover, and a moderator you’ve never met decides it looks generated. What do you do? You can’t prove you didn’t use a tool. Nobody can. You can show sketches and invoices, if you kept them, and hope somebody looks.

Another commenter asked whether there’d be an appeal process for anybody wrongly flagged. That’s the right question. A ban that can’t be appealed is enforced by whoever is in the worst mood that afternoon.

“AI slop” is what people say when they don’t like the picture

I don’t think the pile-on over that ad had anything to do with AI. It was about taste.

You’ve seen it happen. Somebody saw a cover they didn’t care for and reached for the nastiest label available. Ten years ago they’d have said it looked amateur. Today they call it slop, because slop carries an accusation along with the insult. It doesn’t say you lack talent. It says you’re a fraud.

I’ve watched this happen to people restoring their grandmother’s photos and to an author who was called disgusting over a book festival’s cover rule. Platforms are doing it at scale too, with AI labels slapped on human work. The accusation is cheap, it spreads fast and it sticks.

I can’t get past the irony. The loudest anti-AI admin in that thread got burned by the anti-AI crowd. His own side’s favorite insult landed on a painting by his kids’ art teacher. If it can happen to him, it can happen to anybody, and the new rule makes it more likely, not less.

I’ve seen this pattern for years. A crowd picks a villain, and pretty soon anything that looks a little like the villain gets treated as guilty. Covers with smooth lighting. Prose with too many tidy sentences. A writer who’s suddenly more productive. How many honest people have to get hit before anybody admits the test doesn’t work?

What should you do if your book cover gets called AI slop?

Don’t panic. And don’t get into a war in the comments, because you won’t win it.

Say once, plainly, how the cover was made. If you have the artist’s sketches, a photo of the canvas or the designer’s name, post it. That’s all an honest person needs. Anybody who keeps calling you a fraud after that isn’t asking a question, and you don’t owe them a second answer.

If you hired a designer, ask how they work before you pay them. Some designers use AI, and plenty of authors have posted a cover they had no idea was generated. You’d want to know before a moderator decides for you. And keep your files. Your source images, the invoice and the early drafts are the only proof you’ll ever have, so don’t throw them out after launch.

What if you did use AI? Then say so, if the group asks, and follow the group’s rules or post somewhere else. You don’t have to apologize for a tool. You do have to respect a house rule in somebody else’s house.

Do fiction readers care whether an author used AI?

Readers who are writers care.

Readers who are readers mostly don’t.

That’s my opinion, and I’ll stand on it. I’ve been selling my own books for more than a decade. Writers care how a book was made because they make books. They have opinions about tools, process and who did the work, and they argue about them in groups like this one. That’s fair. It’s their trade.

But think about the last time you bought a novel. Did you ask who painted the cover? Most people who buy novels aren’t writers. They want a story that grabs them on page one and doesn’t let go. You glance at a cover for a second or two to decide whether the book is your kind of thing, and so does everybody else. Nobody browsing a thriller is asking what software the designer had open. If the story’s good, they’ll buy the next one. If it isn’t, a hand-painted cover won’t save it.

A lot of the writers in that thread seem to believe readers are as worked up about AI as they are. Are they? A writing group is an echo chamber full of writers, so naturally it sounds like everybody cares. Step outside the group and talk to people who just read, and the subject barely comes up. The writers who care about AI have their reasons, and most of them come back to their own trade. That’s a real worry. It just isn’t the reader’s worry.

His gates, his rules, and the same mistake

One of his moderators said it plainly in the thread: these are their gates, and they’ll keep them. Fine. Keep them. Just keep them fairly.

If you run a group, ask the poster before you delete anything.

Look for the sketches, the source photo, the designer’s invoice. Give your members a way to appeal. And hold the moderators to the same rule the members are held to: no accusations without something better than a feeling. A ban run on vibes will eventually throw out somebody who did everything by hand, the same way those commenters threw out his teacher’s painting.

For more on why AI turns sensible people into prosecutors, see the AI and writing hub and the rest of the Dogpile series. If you want help writing a book that sounds like you, whatever tools you use, my book coaching service is where to start. Judge the book by the book, and let the paint speak for itself.

From the author

Sell Your Books
Sell Your Books starts with identifying your reader before you write to them. Do that and you’ll know whether your buyers care about your tools, or just want a good story and a cover that says what kind of book it is.

Frequently Asked Questions